Effects of antimicrobials on the competitive growth of ... - Oxford Journals

Journal of Antimicrobial Chemotherapy (2006) 58, 794–801 doi:10.1093/jac/dkl307 Advance Access publication 30 July 2006

Effects of antimicrobials on the competitive growth of Streptococcus pneumoniae: a pharmacodynamic in vitro model approach to selection of resistant populations David Sevillano1, Lorenzo Aguilar1, Luis Alou1, Marıá-Jose´ Gimeńez1, Olatz Echevarrıá1, Fabio Cafini1, Eva Valero1, Asuncioń Fenoll2 and Jose´ Prieto1* 1

Microbiology Department, School of Medicine, Universidad Complutense, Avda. Complutense s/n, 28040 Madrid, Spain; 2Centro Nacional de Microbiologıá, Instituto de Salud Carlos III, Ctra. Majadahonda-Pozuelo Km. 2, 28220 Majadahonda, Madrid, Spain Received 4 January 2006; returned 11 May 2006; revised 28 June 2006; accepted 5 July 2006

Objectives: To investigate antimicrobial effects on a mixed culture of five Streptococcus pneumoniae serotypes (S) as an approach to ecology of population dynamics. Methods: A computerized pharmacodynamic model simulating concentrations obtained after levofloxacin, ciprofloxacin and azithromycin doses was used. Resistance patterns were S12, susceptible to study drugs; S31, low-level macrolide-resistant (efflux phenotype); S11, high-level macrolide-resistant (erm genotype); S9V, low-level quinolone-resistant; and S3, high-level quinolone-resistant. Initial mixed inocula (time 0) included similar percentages of each serotype. Results: At 24 h of control drug-free experiments, dominant strains were S9V (57.4%) and S12 (41.8%) with marginal populations of S31, S3 and S11. Azithromycin selected to a much higher extent the strain with lowlevel resistance to macrolides (S31) rather than the strain with high-level resistance (S11) (accounting for 99.9% versus 0.1% of the total population at 24 h). Ciprofloxacin selected to a higher extent lowlevel (S9V) rather than high-level (S3) quinolone resistance (72.4% versus 27.6%). Levofloxacin decreased the proportion of the predominant S9V in controls to 22.2% (an intermediate-resistant strain with MIC = 4 mg/L) and unmasked the high-level resistant strain (MIC = 32 mg/L) up to 77.8%. Conclusions: Strain distribution in an antibiotic-free environment depends on bacterial fitness in monoand multi-strain niches. Selective pressure of antimicrobial regimens eradicate some populations and unmask minor populations, thus redistributing the whole population. Selective potential only for resistance phenotypes with very low prevalence (such as high-level quinolone resistance) in the community should be preferred to that selecting more prevalent resistance phenotypes. Keywords: S. pneumoniae, population dynamics, resistance selection, ecology, pharmacodynamics

Introduction Increasingly, antibiotic resistance is focused as an ecological problem. The pharynx is the reservoir where antibiotic-resistant Streptococcus pneumoniae strains are selected and further spread.1 Normal nasopharynx flora of individuals simultaneously comprise several strains of S. pneumoniae from different serotypes and resistance patterns,2 and at the community level, many individuals harbour multiple pneumococcal populations exhibiting various degrees of antibiotic resistance. A certain dose regimen may select resistant strains despite them constituting a

minority of the total streptococcal population, because antibiotic use seems to be one of the forces behind resistance patterns in different geographic areas rather than clonal dynamics.3–5 The spread of use within a community of a determined antibiotic should produce a shift in the ecology of resistance (both ecology of resistance genes and of resistant bacteria), although little is known about the effect of subinhibitory concentrations (that follow higher concentrations along the dosing interval) on these multi-strain pneumococcal populations in their natural environment. These populations (with different resistance patterns) arise by spontaneous point mutations in the case of

.............................................................................................................................................................................................................................................................................................................................................................................................................................

*Corresponding author. Tel: +34-91-3941508; Fax: +34-91-3941511; E-mail: [email protected] .............................................................................................................................................................................................................................................................................................................................................................................................................................

794 The Author 2006. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved. For Permissions, please e-mail: [email protected]

S. pneumoniae population dynamics quinolone resistance. Antibiotic-resistant populations, regardless of their pathogenic potential, arise by the effect of antibiotics on the nasopharynx, clearing normal flora and allowing their replacement by more resistant strains, or by unmasking and selectively amplifying resistant pre-existing subpopulations.6 In the absence of antibiotics in the environment, the question is whether drug-resistant S. pneumoniae are able to compete with susceptible strains based on the possible different growth rates of resistant and susceptible organisms.7 Dynamics of the different subpopulations in an antibiotic-free environment are the baseline that antibiotic treatments can alter by interfering with bacterial fitness (including colonization and transmission), since resistance can be associated with a decrease in fitness.6 The goal of antibiotic therapy should be both to eradicate the infecting pathogen and to minimize the emergence of resistance. Specific pharmacodynamic parameters based on concentrations achievable in serum during therapy may be used to determine the capability of antibiotic regimens to eradicate susceptible and resistant bacteria and to prevent selection of resistance.6 In vitro pharmacodynamic simulations have previously been used to investigate pharmacodynamic parameters related to bactericidal activity but not to explore S. pneumoniae serotype population dynamics. We report herein an in vitro pharmacodynamic simulation of concentrations obtained after standard doses of three antimicrobials achieving high concentrations (similar or higher than serum) in upper respiratory tract mucosa and saliva,1 and their effect on a mixed culture of five S. pneumoniae serotypes expressing different resistance patterns, as an approach to understand the ecology of S. pneumoniae population dynamics.

Materials and methods Strains Five S. pneumoniae clinical isolates from the Spanish Pneumococcal Reference Laboratory (Instituto de Salud Carlos III, Majadahonda, Madrid, Spain) were selected on the basis of their resistance patterns both with respect to resistance to penicillin, tetracycline, chloramphenicol, erythromycin and levofloxacin (as antibiotics used in selective plates for isolation) and with respect to their resistance to study drugs. Pre- and post-simulation MICs were determined five times by microdilution following CLSI recommendations,8 incubating under 5% CO2, and modal values were considered. Pneumococcal serotyping was carried out by the Quellung reaction using the Statens Serum Institute (Copenhagen, Denmark) typing sera. Table 1 shows the serotypes and resistance patterns of the five strains used. Resistance patterns to the study drugs were S31, low-level resistant to macrolides (efflux phenotype); S9V, low-level

resistant to quinolones; S11, high-level resistant to macrolides (erm genotype); S3, high-level resistant to quinolones; and S12, susceptible to all study drugs.

Antimicrobials Ciprofloxacin and azithromycin laboratory reference standards were supplied by Sigma Chemical Co. (St Louis, MO, USA). Levofloxacin reference standard was supplied by Aventis, S.A., Madrid, Spain.

In vitro kinetic model A previously described two-compartment dynamic model was used to expose bacteria to changing study drug concentrations avoiding the dilution of the bacterial inoculum together with the drug.9 The extracapillary space and the intra-dialyser circulating tubing of the second compartment (FX50 helixone dialyser; Fresenius Medical Care S.A., Barcelona, Spain) represented the colonization site. The central compartment, representing the systemic circulation, consists of a spinner flask with Todd–Hewitt broth (Difco Laboratories, Detroit, MI, USA) supplemented with 0.5% yeast extract (Difco Laboratories), tubing and lumina of capillaries within a dialyser unit. The exponential decay of concentrations was obtained by a continuous dilution–elimination process using computerized peristaltic pumps (Masterflex; Cole-Parmer Instrument Co., Chicago, IL, USA) at a rate of 1 and 1.5 mL/min to simulate half-lives of levofloxacin and ciprofloxacin, respectively, in a distribution volume of 600 mL. For azithromycin, the rate of peristaltic pumps was 3 mL/min in the 2.5–5 h period and 0.47 mL/min in the 5–24 h period, to simulate the biphasic elimination of azithromycin in a distribution volume of 450 mL. These flow rates in the peristaltic pumps were synchronized over all of the simulated period using Win Lin software (Cole-Parmer Instrument Co.). Additional pumps circulated the antimicrobial/ medium mixture at a rate of 50 mL/min rate between the central and peripheral compartment and at 20 mL/min within the extra-capillary space through external tubing. A computer-controlled syringe pump (402 Dilutor Dispenser; Gilson S.A., Villiers-le-Bel, France) allowed the simulation of drug concentrations by infusion of the drug into the central compartment until Cmax was reached. In drug-free simulations, the rate of peristaltic pumps was 1 mL/min. Both compartments were maintained at 37 C over all of the simulation process.

Preparation of individual and mixed cultures The strains were grown overnight on Mueller–Hinton agar supplemented with cations and 5% lysed sheep blood. To prepare the individual inoculum of each strain, a bacterial suspension in 100 mL of Todd–Hewitt broth supplemented with 0.5% yeast extract was

Table 1. Strain ID, serotype and susceptibility (MIC; mg/L) to study drugs (azithromycin, ciprofloxacin and levofloxacin) and to antibiotics used in selective plates (penicillin, tetracycline, chloramphenicol and erythromycin) of strains used in the study Strain ID S31 S9V S11 S3 S12

Serotype

Penicillin

Tetracycline

Chloramphenicol

Erythromycin

Azithromycin

Ciprofloxacin

Levofloxacin

31 9V 11 3 12

0.015 1 0.03 0.015 0.015

16 0.25 0.25 0.5 0.25

4 4 4 4 32

4 0.12 256 0.12 0.06

2 0.03 256 0.06 0.03

1 8 4 128 1

1 4 2 32 0.5

795

Sevillano et al. allowed to grow to a density of 106 cfu/mL, as measured by a UV spectrophotometer (Hitachi U-1100). For the mixed inoculum, a mixture of identical 1:1:1:1:1 volumes was prepared, yielding an inoculum of 5 · 106 cfu/mL.

Experiments Sixty millilitres of each individual inoculum or of the mixed inoculum was introduced into the peripheral compartment of the in vitro model. Experiments performed were (i) antibiotic-free simulations using each individual inoculum (individual growth control); (ii) antibiotic-free simulations using the mixed inoculum (mixed growth control); and (iii) simulations with each antimicrobial drug using the mixed inoculum. In each experiment, samples (0.5 mL) from the peripheral compartment were collected at 0, 1, 2, 4, 6, 8, 10, 12 and 24 h. Each sample was serially 10-fold diluted in 0.9% sodium chloride for bacterial counting onto plates containing supplemented Mueller– Hinton agar with 5% sheep blood and with different antibiotics to identify the different strains over time. Plates containing 8 mg/L of tetracycline were used for selective growth of strain S31, those containing 0.5 mg/L of penicillin for strain S9V, those containing 64 mg/L of erythromycin for strain S11, those containing 8 mg/L of levofloxacin for strain S3 and those containing 8 mg/L of chloramphenicol for strain S12. In addition, antibiotic-free plates were used for global counting both with individual and with mixed inocula. All plates were incubated at 37 C in 5% CO2 for 24 h. The limit of detection was 50 cfu/mL. All experiments were performed in triplicate. Ten colonies from cultures obtained in antibiotic-free plates after 24 h simulations (performed with antimicrobials) in experiments carried out with the mixed inoculum were isolated and serotyped.

Kinetic simulations Pharmacokinetic profiles in serum after oral single doses of azithromycin 500 mg once daily, ciprofloxacin 750 mg twice daily and levofloxacin 500 mg once daily were simulated over 24 h. The target pharmacokinetic parameters, based on mean values reported in humans, were Cmax = 5.63 mg/L and t1/2 = 6.8 h for levofloxacin10–13 and Cmax = 3.3 mg/L and t1/2 = 4.5 h for ciprofloxacin.14,15 The target serum concentrations of azithromycin were obtained from human serum concentration-versus-time curves16,17 because serum concentrations of azithromycin decline in a polyphasic manner, and its relatively short serum half-life between 8 and 24 h after dosing indicates its initial rapid tissue distribution. Experimentally, the profile was divided into two quasi-linear portions with apparent half-lives of 1.71 h (Ke; 0.4 h–1) from 2.5 to 5 h and of 10.5 h (Ke; 0.06 h–1) from 5 to 24 h representing further distribution and elimination. Final target pharmacokinetic parameters were Cmax = 0.4 mg/L and t1/2 = 10.5 h.

Pharmacokinetic analysis For the measurement of simulated antimicrobial concentrations, aliquots (0.5 mL) were taken at 0, 1, 1.5, 2, 2.5, 4, 6, 8, 10, 12 and 24 h and stored at –50 C. Concentrations were determined by bioassay using Escherichia coli NCTC 10418 for levofloxacin and ciprofloxacin18 and Micrococcus luteus ATCC 9341 for azithromycin.19 Plates were inoculated with an even lawn of indicator organism and incubated for 18–24 h at 37 C. Standards were prepared in Todd–Hewitt broth supplemented with 0.5% yeast extract, with a concentration range of 0.06–4 mg/L for ciprofloxacin and levofloxacin and of 0.06–0.5 mg/L for azithromycin. Lower limits of

detection were 0.06, 0.03 and 0.06 mg/L for levofloxacin, ciprofloxacin and azithromycin, respectively. Intra-day coefficients of variation were 3.1%, 2.6% and 6.2% while inter-day coefficients of variation were 2.8%, 0.7% and 5% for levofloxacin, ciprofloxacin and azithromycin at concentrations of 0.75, 0.75 and 0.1 mg/L, respectively. Antimicrobial concentrations were analysed by a noncompartmental approach using the WinNonlin Professional program (Pharsight, Mountainview, CA, USA). Cmax and Tmax were obtained directly from observed data. The area under the concentration-time curve from 0 to 24 (AUC0–24) was calculated by the trapezoidal rule.

Data analysis After individual antibiotic-free simulations, colony counts in selective media were compared with those in antibiotic-free plates to determine the efficacy of the selective plates in strain recovery. In experiments carried out with the mixed inocula, initial inocula of each strain was the mean value of the 12 determinations performed, and the coefficient of variation was calculated. After simulations carried out with the mixed inoculum, differences between colony counts in antibiotic-free plates (whole pneumococcal population) and the sum of individual counts in selective plates were calculated. These differences, due to pneumococcal populations not detectable in selective plates, were defined as the ‘non-detectable population’ (NDP) and expressed as the percentage of total pneumococcal population in mixed inocula experiments. At 24 h, NDPs were unmasked in antibiotic-free simulations with the mixed inoculum by extrapolation of the specific weight of each strain in mixed cultures [assessed as the percentage in selective plates versus whole pneumococcal population (antibiotic-free plates)]. These values were corrected considering the percentage at 24 h of NDP for each strain obtained in individual antibiotic-free simulations when comparing selective versus non-selective plates. NDPs in antimicrobial simulations with the mixed inoculum were unmasked by the combination of two factors: morphology (mucous colonies of strain S3, confirmed by serotyping) and serotyping of smooth colonies obtained at 24 h in antibiotic-free plates. Once all populations were unmasked, results were expressed as the percentage of each strain with respect to the whole pneumococcal population in mixed inocula experiments. Comparisons of growth in selective plates of each strain in individual versus mixed drug-free simulations were performed using the Student’s t-test. P < 0.05 was considered significant.

Results Pharmacokinetics Figure 1 shows target and experimental concentrations of the antimicrobials used in the study. Experimental pharmacokinetic parameters for levofloxacin, ciprofloxacin and azithromycin were Cmax (mg/L), 5.7 – 0.4, 3.2 – 0.2 and 0.4 – 0.04; Tmax (h), 1.4 – 0.4, 1.4 – 0.3 and 2.5 – 0.0; and half-life (h), 6.7 – 0.3, 4.8 – 0.2 and 11.1 – 0.7, respectively. AUC0–24 (mg · h) values were 52.8 – 2.3, 18.3 – 0.6 and 2.7 – 0.2, respectively.

Drug-free simulations Figure 2 (a–e) shows colony counts in selective media compared with those in antibiotic-free plates after individual drug-free simulations for each strain. While no significant differences in colony counts over the 24 h simulation period were found for

796

S. pneumoniae population dynamics 7

Levofloxacin 500 mg once daily Ciprofloxacin 750 mg twice daily Azithromycin 500 mg once daily

6 Concentration (mg/L)

5 4 3 2 1 0.4 0.3 0.2 0.1 0.0 0

2

4

6

8

10

12 14 Time (h)

16

18

20

22

24

Figure 1. Target (dashed lines) and experimental concentrations of study drugs.

Strain S31

0.99 ± 0.99 log10 cfu/mL (9.36 × 108 cfu/mL)

8 6 4 2 4

8

(c)

12 16 Time (h)

20

24

Strain S11

10 8 6

0.99 ± 0.77 log10 cfu/mL (1.76 × 105 cfu/mL)

4 2

0

4

8

(e)

12 16 Time (h)

20

24

10 1.33 ± 0.36 log10 cfu/mL (6.50 × 108 cfu/mL)

8 6 4 2 8

12 16 Time (h)

0.12 ± 0.08 log10 cfu/mL (7.40 × 108 cfu/mL)

8 6 4 2 0

4

8

12 16 Time (h)

20

24

Strain S3 12 10

0.01 ± 0.02 log10 cfu/mL (1.50 × 107 cfu/mL)

8 6 4 2 0

4

(f)

Strain S12

4

10

8

12

16

20

24

Time (h)

12

0

12

(d)

12

Viable counts (log10 cfu/mL)



10

0

Strain S9V

(b)

12



(a)


strains S9V and S3, lower colony counts in selective plates were obtained over time for strains S31, S11 and S12. Thus, high NDPs (versus the total population) were found in experiments with strains S31, S11 and S12. On the contrary, low NDPs were found with strains S9V and S3. All strains but S11 grew from time 0 onwards, maintaining high colony counts over time regardless of the use or not of selective plates. Strain S11 started to decrease after 8 h. Figure 2(f) shows colony counts on selective media of each strain and the whole population determined on antibiotic-free plates after drug-free simulations using the mixed inoculum. Antibiotic-free plates yielded higher colony counts than the sum of bacterial counts on selective plates. When comparing in drugfree simulations the growth on selective plates after individual experiments (Figure 2a–e) with the growth on selective plates after mixed inocula experiments (Figure 2f), strains S31 and S3 showed significantly (P < 0.05) lower colony counts from 6 h onwards in experiments with the mixed inocula versus the individual ones.

20

24

Mixed control S31

12

S9V

10

S11 S3

8

S12 Total population in antibiotic-free plates

6 4 2 0

4

8

12 16 Time (h)

20

24

Figure 2. Colony counts in selective media versus antibiotic-free plates (dashed lines) for each strain after individual (a–e) and mixed (f) drug-free simulations. Differences between colony counts in antibiotic-free plates and in selective media at 24 h are shown (expressed as differences in cfu/mL and differences in log10 cfu/mL).

797

Sevillano et al. Simulations with study drugs and mixed inocula Mean colony counts on antibiotic-free plates (whole pneumococcal population) increased (from 0 to 24 h) from log10 6.9 to 8.8 in drug-free simulations (control), from log10 6.8 to 7.9 in levofloxacin simulations, from log10 6.8 to 8.6 in ciprofloxacin simulations, and from log10 7.1 to 8.8 in azithromycin simulations. The coefficient of variation of the percentage of each serotype in the initial mixed inocula ranged from 0.18% for strain S9V to 0.48% for strain S11. Figure 3 shows the evolution of mixed inocula over time (expressed as the percentage of growth on selective plates and % NDPs) in experiments with and without study drugs. The first five columns correspond to the population of the five serotypes plus the NDP at 0, 4, 8, 12 and 24 h. The last column shows the final serotype distribution (percentage of the five serotypes after unmasking the NDPs). At time 0, mean population distribution of the strains within the mixed inocula ranged from 11.6% for S12 to 19.6% for S9V. At the end (24 h) of control drug-free experiments, dominant strains were S9V (57.4%) and S12 (41.8%) with 0.5% of S31, 0.2% of S3 and 0.1% of S11. At the end of levofloxacin experiments, the dominant strain was S3 (77.8%) followed by S9V (22.2%). This figure is inverted at the end of ciprofloxacin experiments where the dominant strain was S9V (72.4%) followed by S3 (27.6%). Finally, at the end of azithromycin simulations, the dominant strain was S31 (99.9%) with a marginal population of S11 (0.1%). MICs for all strains exhibited the same values before and after the simulation processes.

Pharmacodynamics Table 2 shows the pharmacodynamic parameters of study drugs for the five strains used in the study. With respect to the initial mixed inocula, antibacterial activity of study drugs was observed with high AUC0–24/MIC values, except in the case of S11 where both ciprofloxacin and azithromycin parameters had low values. With respect to the figure of controls at 24 h, levofloxacin eradicated S12 and widely reduced the S9V percentage, highly increasing the S3 percentage. Ciprofloxacin eradicated S12, increasing S3 (to a lesser extent than levofloxacin) and S9V percentages. Azithromycin eradicated S12 and S9V, highly increasing the S31 percentage (from 0.5 to 99.9%) and maintaining the S11 percentage (0.1%).

Discussion Ecology and resistance in human microbiota are closely related since the second phenomenon is a marker of the first, and it is assumed with some evidence, at least at short term, that bacteria pay a physiological price (decrease in fitness) for resistance.20,21 Little is know about fitness of antimicrobialresistant S. pneumoniae,7 a species with different serotypes that can be simultaneously carried by the same individual. Usually one strain is predominant, and minor S. pneumoniae populations (with specific resistance patterns) that can be simultaneously carried by multiple individuals in the community cannot be detected unless a large number of colonies are analysed,22 or if these resistant minor populations are selected by a certain antibiotic dose regimen.2 Considering this, serotypes isolated in

non-antibiotic-treated humans may differ from those isolated in antibiotic-treated patients due to three facts: detection, fitness and antimicrobial unmasking of minor populations.

Detection With respect to detection, selective media can be used to discriminate different strains from the same sample provided they have different resistance patterns. But the sensitivity of this method depends on the subpopulation distribution of resistance within a single strain and probably on the antibiotic and the antibiotic concentration present in the selective media. In this sense, in our study, strains S9V and S3 exhibited in individual drug-free simulations lower differences in colony counts between antibiotic-free plates and selective media (penicillin for S9V and levofloxacin for S3) than the other strains. On the other hand, when tetracycline, chloramphenicol or erythromycin was used in selective media lower bacterial counts were obtained in comparison with those in antibiotic-free plates for strains S31, S11 and S12. Thus, there are subpopulations within a specific strain that in our study cannot be detected.

Fitness With respect to bacterial fitness (defined as differential growth rate between susceptible and resistant strains), in individual experiments of the study, strains with low- and high-level quinolone resistance (strains S9V and S3, respectively) and the strain with low-level azithromycin resistance (strain S31) showed over 24 h an increase in bacterial counts similar to that of the susceptible strain (S12). In contrast, the strain with high-level resistance to erythromycin (strain S11) suffered autolysis from 8 h onwards. Other authors have reported that mutations in parC and gyrA genes, on some occasions, are not associated with a physiological deficit.21 When measuring bacterial fitness as the capability of growth in drug-free simulations with the mixed inocula, strains S9V, S11 and S12 exhibited similar growth rates in experiments with the individual and the mixed inocula (when detected in their respective selective plates), whereas strains S31 and S3 showed statistically significant lower colony counts over time when grown in mixed inocula. The decrease of these two strains when they are accompanied by other serotypes cannot be attributed to a competition for nutrients because fresh medium was continuously supplied in the simulation device. The capability of utilization of nutritional elements may be the clue for these differences. Changes in serotype proportion in the mixed inocula over time in drug-free simulations are the consequence of each serotype’s fitness (both in individual and in mixed experiments). This is of paramount importance in analysing the effects of antimicrobials on S. pneumoniae ecology. In the present study, from a similar percentage of each serotype at time 0, the best global fitness corresponded to S9V and S12 that together represent 99.2% of the final population at 24 h after identification of NDPs. The other three serotypes constituted minor (0.05%) pneumococcal populations that would be difficult to detect in field studies.

Antimicrobial unmasking of minor populations Antimicrobial drugs selectively unmask minor populations, thus changing the ecology (serotype distribution) of an antimicrobialfree environment, primarily by eradicating susceptible strains.

798

799 S31; 14.4

NDP; 18.0

S11; 21.9

Azithromycin 500 mg

S3; 0.2

NDP; 77.3

S31; 0.6

S9V; 3.1

NDP; 13.9

Ciprofloxacin750 mg

S11; 0.1

NDP; 2.9

S9V; 0.5

Control (Placebo)

S3; 19.5 S11; 6.8

Levofloxacin 500 mg

S3; 82.9

S3; 96.6

S12; 1.2

S31; 2.2 S9V; 5.6

NDP; 64.7

4h

NDP; 86.5

S3; 70.9

S3; 70.5

S31; 2.6

S11; 4.3

S31; 9.2

S9V; 0.4

NDP; 28.7

NDP; 29.5

S3; 11.6 S11; 0.5 S9V; 23.8

S12; 7.8

NDP; 53.7

8h

S31; 0.6

S31; 9.7

S9V; 0.7

NDP; 29.1

NDP; 42.5

S9V; 33.7

NDP; 55.1

S11; 1.4

NDP; 88.9

S3; 70.2

S3; 57.5

S11; 0.2

S3; 7.9

S12; 2.5

12 h

Figure 3. Evolution of mixed inocula over time with and without study drugs [non-detectable populations (NDPs) are in white].

S9V; 19.6

S11; 18.2

S3; 18.1

S12; 11.6

Initial distribution 0 h

NDP; 98.4

S3; 77.8

S3; 0.2

S11; 0.1

S3; 27.6

S31; 1.5

S9V; 0.1

NDP; 22.2

S9V; 46.7

NDP; 52.0

NDP; 72.3

S12; 1.0

24 h

S31; 99.9

S9V; 72.4

S3; 77.8

S9V; 57.4

S11; 0.1

S3; 0.2

S31; 0.5

S12; 41.8

S11; 0.1

S3; 27.6

S9V; 22.2

Final serotype distribution

S. pneumoniae population dynamics

Sevillano et al. Table 2. Pharmacodynamic parameters Levofloxacin Strain ID S31 S9V S11 S3 S12

Ciprofloxacin

Azithromycin

AUC/MIC

Cmax/MIC

T > MIC (%)

AUC/MIC

Cmax/MIC

T > MIC (%)

AUC/MIC

Cmax/MIC

T > MIC (%)

52.8 13.2 26.4 1.7 105.6

5.7 1.4 2.8 0.2 11.3

78.8 15 43.8 0.0 99.6

18.3 2.3 4.6 0.1 18.3

3.2 0.4 0.8 0.0 3.2

74.6 0.0 0.0 0.0 74.6

1.4 90.7 0.0 45.3 90.7

0.2 12.7 0.0 6.3 12.7

0.0 99.2 0.0 88.1 99.2

In our study S12, which represented 41.8% of the pneumococcal population after 24 h drug-free simulations, was driven to values under the limit of detection by the three study drugs, despite nonhigh AUC0–24/MIC values of ciprofloxacin. However, against this strain, T > MIC values were 74% of the dosing interval with the three compounds. Azithromycin selected (unmasked) the minor S31 population that accounted for 0.5% in controls, but for 99.9% after azithromycin simulation. Interestingly, S31 exhibited lower growth ability when accompanied by other serotypes (lower fitness in mixed inocula). When azithromycin drove the strains with best fitness in mixed inocula (S9V and S12), and the minor population of S3 in controls to values under the detection limit, S31 regained its growth ability. The remaining population after azithromycin treatment was S11, which was maintained at similar levels as in controls (0.1%). S31 presented an efflux phenotype of low-level resistance to macrolides, while strain S11 exhibited a constitutive phenotype of high-level resistance. Azithromycin selected to a much higher extent the strain with low-level resistance to macrolides (S31) than the strain with high-level resistance (S11) that presented poor fitness in individual experiments. These facts could help to explain the increase in the prevalence of the M phenotype among resistant strains in successive surveillance studies in our country23,24 and the major spread of resistance due to this phenotype in many other parts of the world.25 In ciprofloxacin simulations the well-fitted population of S9V in controls increased from 57.4% to 72.4%, unmasking the minor population of S3 that increased from 0.2% to 27.6%, due to the eradication of the other major well-fitted serotype in controls (S12). Ciprofloxacin selected to a higher extent low-level than high-level quinolone resistance. This may have implications in treatments in the community since low-level resistance is more prevalent,23–27 and ciprofloxacin has the highest consumption rate among quinolones.4 Because ciprofloxacin is more extensively used as treatment of uncomplicated urinary tract infections, treatments directed to community uropathogens may be responsible for the selection of low-level quinolone resistance in S. pneumoniae in the pharynx. This is consistent with the results observed by Chen et al.,28 who reported an increase in fluoroquinolone-resistant S. pneumoniae in Canada associated with an increase in fluoroquinolone usage. Ciprofloxacin was the predominant fluoroquinolone used during the study period. On the other hand an inverse correlation between quinolone consumption and non-susceptibility to ciprofloxacin by province was found in Spain29 in part due to the influence of the replacement of ciprofloxacin by quinolones (such as levofloxacin), with much

higher activity against S. pneumoniae, that have reduced selection towards higher rates of resistance.29 More than 85% of S. pneumoniae isolates resistant to ciprofloxacin (MIC 4 mg/L) are susceptible to levofloxacin27 and for this reason non-susceptibility rates to levofloxacin are 1%,30 with very low prevalence of high-level resistance to levofloxacin (as in the strain used in our experiment) reported.27 In our experiments, levofloxacin eradicated susceptible serotypes (S31, S11 and S12) due to the high AUC0–24/MIC, decreased the proportion of the predominant S9V in controls from 57.4% to 22.2% (an intermediate-resistant strain with an MIC of 4 mg/L) and unmasked, as expected, the high-level resistant strain (MIC of 32 mg/L). Pneumococcal epidemiology in an antibiotic-free environment depends on bacterial fitness both in mono- and multi-strain ecological niches, resulting in a serotype distribution over time with predominant and minor populations, for which detection depends on technical issues. The selective pressure of antibiotic treatments eradicates some populations and unmasks minor populations, thus redistributing the whole population. By using resistance as a marker of ecology, the selective effect of each antibiotic can be measured. Antibiotics that only select non-prevalent resistance phenotypes in the community should be preferred to those selecting prevalent phenotypes. In the case of quinolones where resistance can arise from selection of resistant strains or from point mutations in a single strain population, resistance development with levofloxacin is also less frequent than with ciprofloxacin.31 Further studies, maybe using this model approach but with different parameters (serotypes, antibiotics and dosing regimens), could be useful for progressing the study of population dynamics of S. pneumoniae within human microbiota.

Acknowledgements We thank J. E. Martıń and C. Garcıá-Rey for their critical review of the manuscript prior to submission. This study was supported in part by an unrestricted grant from GlaxoSmithKline S.A., Madrid, Spain and Sanofi-Aventis S.A., Barcelona, Spain.

Transparency declarations None to declare.

References 1. Soriano F, Rodriguez-Cerrato V. Pharmacodynamic and kinetic basis for the selection of pneumococcal resistance in the upper respiratory tract. J Antimicrob Chemother 2002; 50 Suppl S2: 51–8.

800

S. pneumoniae population dynamics 2. Knudsen JD, Odenholt I, Erlendsdottir H et al. Selection of resistant Streptococcus pneumoniae during penicillin treatment in vitro and in three animal models. Antimicrob Agents Chemother 2003; 47: 2499–506. 3. Garcia-Rey C, Fenoll A, Aguilar L et al. Effect of social and climatological factors on antimicrobial use and Streptococcus pneumoniae resistance in different provinces in Spain. J Antimicrob Chemother 2004; 54: 465–71. 4. Goossens H, Ferech M, Vander Stichele R et al. Outpatient antibiotic use in Europe and association with resistance: a crossnational database study. Lancet 2005; 365: 579–87. 5. McCormick AW, Whitney CG, Farley MM et al. Geographic diversity and temporal trends of antimicrobial resistance in Streptococcus pneumoniae in the United States. Nat Med 2003; 9: 424–30. 6. Nuermberger EL, Bishai WR. Antibiotic resistance in Streptococcus pneumoniae: what does the future hold? Clin Infect Dis 2004; 38 Suppl 4: S363–71. 7. Johnson CN, Briles DE, Benjamin WH, Jr et al. Relative fitness of fluoroquinolone-resistant Streptococcus pneumoniae. Emerg Infect Dis 2005; 11: 814–20. 8. Clinical and Laboratory Standards Institute. Performance Standards for Antimicrobial Susceptibility Testing: Fifteenth Informational Supplement M100-S15., CLSI, Wayne, PA, USA, 2005. 9. Sevillano D, Calvo A, Gimenez MJ et al. Bactericidal activity of amoxicillin against non-susceptible Streptococcus pneumoniae in an in vitro pharmacodynamic model simulating the concentrations obtained with the 2000/125 mg sustained-release co-amoxiclav formulation. J Antimicrob Chemother 2004; 54: 1148–51. 10. Chien SC, Rogge MC, Gisclon LG et al. Pharmacokinetic profile of levofloxacin following once-daily 500-milligram oral or intravenous doses. Antimicrob Agents Chemother 1997; 41: 2256–60. 11. Fish DN, Chow AT. The clinical pharmacokinetics of levofloxacin. Clin Pharmacokinet 1997; 32: 101–19. 12. Lee LJ, Hafkin B, Lee ID et al. Effects of food and sucralfate on a single oral dose of 500 milligrams of levofloxacin in healthy subjects. Antimicrob Agents Chemother 1997; 41: 2196–200. 13. Lubasch A, Keller I, Borner K et al. Comparative pharmacokinetics of ciprofloxacin, gatifloxacin, grepafloxacin, levofloxacin, trovafloxacin, and moxifloxacin after single oral administration in healthy volunteers. Antimicrob Agents Chemother 2000; 44: 2600–3. 14. Israel D, Gillum JG, Turik M et al. Pharmacokinetics and serum bactericidal titers of ciprofloxacin and ofloxacin following multiple oral doses in healthy volunteers. Antimicrob Agents Chemother 1993; 37: 2193–9. 15. Shah A, Lettieri J, Blum R et al. Pharmacokinetics of intravenous ciprofloxacin in normal and renally impaired subjects. J Antimicrob Chemother 1996; 38: 103–16. 16. Lode H. The pharmacokinetics of azithromycin and their clinical significance. Eur J Clin Microbiol Infect Dis 1991; 10: 807–12. 17. Capitano B, Mattoes HM, Shore E et al. Steady-state intrapulmonary concentrations of moxifloxacin, levofloxacin and azithromycin in older adults. Chest 2004; 125: 965–73.

18. Andrews JM. Microbiological assays. In: Reeves DS, Wise R, Andrews JM, White LO, eds. Clinical Antimicrobial Assays. 1st edn. Oxford: Oxford University Press, 1999; 35–44. 19. Bonnet M, Van der Auwera P. In vitro and in vivo intraleukocytic accumulation of azithromycin (CP-62, 993) and its influence on ex vivo leukocyte chemiluminescence. Antimicrob Agents Chemother 1992; 36: 1302–9. 20. Gillespie SH. Antibiotic resistance in the absence of selective pressure. Int J Antimicrob Agents 2001; 17: 171–6. 21. Gillespie SH, Voelker LL, Dickens A. Evolutionary barriers to quinolone resistance in Streptococcus pneumoniae. Microb Drug Resist 2002; 8: 79–84. 22. Schrag SJ, Beall B, Dowell SF. Limiting the spread of resistant pneumococci: biological and epidemiologic evidence for the effectiveness of alternative interventions. Clin Microbiol Rev 2000; 13: 588–601. 23. Baquero F, Garcia-Rodriguez JA, Garcia de Lomas J et al. Antimicrobial resistance of 1,113 Streptococcus pneumoniae isolates from patients with respiratory tract infections in Spain: results of a 1-year (1996–1997) multicenter surveillance study. Antimicrob Agents Chemother 1999; 43: 357–9. 24. Perez-Trallero E, Garcia-de-la-Fuente C, Garcia-Rey C et al. Geographical and ecological analysis of resistance, coresistance, and coupled resistance to antimicrobials in respiratory pathogenic bacteria in Spain. Antimicrob Agents Chemother 2005; 49: 1965–72. 25. Ambrose KD, Nisbet R, Stephens DS. Macrolide efflux in Streptococcus pneumoniae is mediated by a dual efflux pump (mel and mef) and is erythromycin inducible. Antimicrob Agents Chemother 2005; 49: 4203–9. 26. Balcabao IP, Alou L, Aguilar L et al. Influence of the decrease in ciprofloxacin susceptibility and the presence of human serum on the in vitro susceptibility of Streptococcus pneumoniae to five new quinolones. J Antimicrob Chemother 2001; 48: 907–9. 27. Perez-Trallero E, Garcia-Rey C, Martıń-Sańchez AM et al. Activity of six different quinolones against clinical respiratory isolates of Streptococcus pneumoniae with reduced susceptibility to ciprofloxacin in Spain. Antimicrob Agents Chemother 2002; 46: 2665–7. 28. Chen DK, McGeer A, de Azavedo JC et al. Decreased susceptibility of Streptococcus pneumoniae to fluoroquinolones in Canada. N Engl J Med 1999; 341: 233–9. 29. Garcia-Rey C, Martin-Herrero JE, Baquero F. Antibiotic consumption and generation of resistance in Streptococcus pneumoniae: the paradoxical impact of quinolones in a complex selective landscape. Clin Microbiol Infect 2006; 12 Suppl 3: 55–66. 30. Barberań J, Gimeńez MJ, Aguilar L et al. Scientific evidence and global conception of empirical treatment of lower respiratory tract infections in the community. Rev Esp Quimioter 2004; 17: 317–24. 31. Madaras-Kelly KJ, Demasters TA. In vitro characterization of fluoroquinolone concentration/MIC antimicrobial activity and resistance while simulating clinical pharmacokinetics of levofloxacin, ofloxacin, or ciprofloxacin against Streptococcus pneumoniae. Diagn Microbiol Infect Dis 2000; 37: 253–60.

801